►
From YouTube: Kubernetes SIG CLI 20191218
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
There
initial
comments
that,
instead
of
just
using
the
ephemeral
containers,
we
can
build
a
much
future
full
command
with
either
being
the
user
could
be
used
either
picking
whether
he
wants
to
use
a
ephemeral
containers
or
a
regular
pod.
I
do
hope
that
we
will
be
able
to
get
agreement
on
the
proposal.
It
was
discussed
heavily
in
the
past
meetings,
so
we
have
an
agreement
on
the
shape
of
it.
A
B
Yeah
I
just
left
a
comment
that
we're
building
out
like
surface
area
and
customize
for
commands
that
Michael,
that
would
otherwise
be
in
coop
control,
but
we
can
iterate
faster
on
and
then
we
have
a
simpler
packaging
story
than
the
plugins
I
think
so
we
may
want
to
add
it
there
as
well
I'll
talk
about
more
about
that
at
the
end.
Here,
okay,.
A
Okay,
yeah,
that's
that's
one
option
or
we
can
still
yeah.
Let's
talk
it
through
further
on.
Does
anyone
have
any
questions
with
regards
to
this
particular
item
that
we
were
working
on
during
the
118
timeframes,
if
people
are
interested
in
helping
both
with
the
proposal
or
implementation
just
reach
out
to
me
orally
and
hopefully
we'll
be
able
to
direction
the
right
in
the
right
place?
A
I
do
see,
one
should
put
together
a
a
proposal
for
for
the
new
command
again
additional
new
command
for
the
events
at
which
that
will
be
much
more
feature.
Phul
than
the
current
get
the
current
get
is
very
limited.
You
can
get
either
all
events
or
events
from
a
particular
namespace
like
with
that
any
regular
get
resource
where,
as
we
will
want
to
have
some
more
filtering
and
sorting
capabilities
added
into
that
command.
All
the
details.
A
Okay,
so
the
next
is
our
long
lived
process
that
Shawn
and
big
thanks
to
Shawn
for
doing
this.
For
the
past
couple
of
months,
Shawn
is
driving
the
extraction
of
cute
you've
got
all
to
staging
we're
pretty
much
seeing
the
light
and
the
tunnel
there
are
me.
Thank
you,
leftovers
Shawn.
Can
you
give
a
quick
update
where
you
are,
how
far
we
are
I?
Remember
the
one
of
them.
Mainly,
offenders
is
authorizing
style.
I
had
a
discussion
with
Jordan
during
cute
con.
We
hadn't,
given
him
that
either
I'm
gonna
continue
the
work
that.
A
The
other
option
is
I'm
gonna
copy
the
code
and
ensure
that
the
code
that
will
be
in
Cuba,
caudal,
staging
and
within
the
main
repo
does
not
diverge.
Whichever
solution
will
be
faster
for
me
to
finish,
we
will
go
with
that
and
that's
the
agreement
that
I
have
with
Jordan
I'll
try
to
go
with
the
library
approach,
because,
obviously
the
the
library
approach
would
be
the
long
term,
the
other,
although
seems
simpler,
it
will
have
a
limited
life
cycle
because
at
some
point
in
time,
we'll
have
to
eventually
fix
the
the
library
code.
D
Yeah
so
there's
there's
two
commands
subcommittee
kook
little
sub
commands
that
haven't
been
pulled
out
of
kubernetes
core.
The
first,
as
you
mentioned,
is
the
coop
cuddle
auth
reconcile
and
the
there's
one
dependency
on
the
our
back,
which
is
exactly
what
would
much
Hays
been
discussing
and
we
believe
the
the
answer.
The
long-term
answer
is
to
pull
that
dependency
into
its
own
posit
Ori,
so
that
can
be
reused,
but
so
so,
in
addition,
it
would
be
outside
of
kubernetes
core.
D
The
second
is
coop
cuttle
convert
which,
because
it
converts
and
and
conversions,
go
through
internal
objects
or
internal
versions
of
objects
that
convert
can
never
be
fixed,
and
so
our
our
resolution
for
that
is
that
we're
going
to
create
a
a
second
created
as
a
standalone
binary,
so
that
the
functionality
will
still
exist.
However,
it
won't
be
in
coop
control,
it'll,
be
probably
a
binary
called
convert
or
something
similar.
If
you've
got
opinions
on
that,
please
let
me
know
I've
already
gotten
something
along
the
lines
of
that
working.
D
It's
pretty
straightforward.
We
just
need
to
figure
out
kind
of
what
the
flags
or
some
of
the
very
simple
interface
stuff
is
going
to
work,
but
that's
that's
all
we
have
left.
Actually,
that's
not
one
ever
said
true.
We
have
a
couple
of
other
pieces
that
are
still
in
kubernetes
core
that
can
easily
be
moved
out
once
we
have
those
two
pieces
out,
specifically
coop
kuddle
copy,
which
was
moved
back
into
core
because
of
the
security
problems
with
it.
D
E
This
is
Julian,
so
we
merged
the
kept
for
the
plan
to
use
the
dry
run
flag
for
a
cube
tunnel.
Yeah
thanks
thanks
for
your
help
for
making
that
the
plan
is
to
start
explicitly
warning
users
that
you
must
now
set
a
value
for
dry
run.
So
currently
you
don't
need
to
set
a
value
for
dry
run,
but
that
would
allow
us
to
get
to
a
pass
where
users
would
explicitly
set
dry,
run,
equals
client
or
drying
equals
server
or
the
type
of
dry
run.
So
we're
gonna
work
on
that
and
I
think.
A
I've
noticed
you
had
a
PR
open
and
it
consists
of
close
to
40
or
more
than
40
comments,
or
it
was
38
comments
and
44
files
changed.
Do
you
think
you
can
split
this
split
us
out
into
a
smaller
pieces
so
that
it'll
be
easier
for
review?
I
started
looking
at
it,
but
I
I
freaked
out
a
little
bit
because
of
the
size.
It
is
definitely
easier
for
me
to
there's
spend
time
on
smaller
bits
as
young
as
they
make
sense
together
and
they
don't
break
the
integrity
of
the
functionality.
You
I'm
perfectly
fine,
yeah.
D
F
A
Cool
the
next
topic-
I
put
it
here
as
a
tension
up
sure
if
or
how
far
we
get
with
it.
I
have
an
initial
agreement
from
the
sig
API
machinery,
folks,
with
regards
to
tracking
the
headers
of
tracking
the
commands
that
are
being
invoked
in
the
server
so
with
the
overall
approach
that
we
are
trying
to
increase
them.
A
Monitoring,
capabilities
of
the
inherent
kubernetes
cluster
Phil
put
together
a
proposal,
a
couple
ones
back,
which
is
discussing
the
ability
to
inject
the
commands
and
some
parameters
from
the
commands
and
the
proposal,
and
unless
which
specifically
would
on
a
C
and
then
they
will
be
sent
over
to
the
server
and
then
the
server
can
create
the
metrics
and
the
metrics
based
on
those
metrics.
We
could
then
get
a
feedback
which
commands
are
being
used
and
what
kind
of
parameters
are
people
most
frequently
passing
over
to
them?
A
A
Gathering
direct
feedback
from
users,
either
through
surveys
or
something
similar,
which
also
reminds
me
that
the
six
user
experience
we'll
be
conducting
a
survey
about
the
overall
excuse.
Our
experience
with
gratis
I
want
to
sync
with
them
with
regards
to
the
cute
cuddle
part
and
gets
the
results,
at
least
in
the
area,
that
we
are
interested
in,
not
sure
how
far
we
like
I'm
saying,
I'm,
not
sure
how
far
we
will
be
able
to
get
within
118.
A
B
This
yeah,
this
would
also
be
it
would
be
great
to
get
us
some
data,
there's
also
various
things
that
we
can
do.
That
would
be
interesting
with
this,
such
as
we
can
for
committing
to
a
deprecating.
We
can
send
as
part
of
the
header
that
a
deprecated
command
is
being
used
and
then
the
api
sort,
like
a
user,
could,
in
theory,
stall
a
service
into
their
cluster.
That
could
look
at
the
logs
to
see
if
they're,
using
deprecated
commands
and
then
have
like
cluster
admins,
be
aware
of
this
or
so
like
allow
organizations.
B
A
B
Yeah,
so
we
started
building
a
number
of
commands
in
customized.
On
top
of
this
set
of
libraries
called
KML.
B
Libraries
in
these
commands
are
the
idea
is
that,
like
these
are
useful,
CLI
commands
and
instead
of
putting
them
in
coop
control,
which
has
a
long
release
cycle,
and
it
just
has
a
lot
more
going
on
like
there's
a
much
faster
iteration
cycle
in
customize,
you
can
just
go,
get
it
and
so
they're
enabled,
with
a
flag
or
sorry
an
environment
variable
to
enable
the
alpha
commands
and
if
you
run
customize
you'll
see
them
play
this
version
of
customize
and
then
I
can
give
a
demo
at
the
end
of
one
of
the.
B
B
G
You're
following
kind
of
a
UNIX
philosophy
for
creating
small
composable,
client-side
programs
or
manipulating
declarative
configurations
and
kind
of
the
idea
here
is
that
customized
config
run
would
be
a
reference
implementation
of
a
reference
implementation
for
orchestration
of
these
configuration
functions,
but
we
could
implement
both
the
functions
and
orchestrations
in
any
tool
chain
or
any
framework
or
any
language.
So,
if
you're
interested,
please
take
a
look
and
I
love
to
hear
comments
on
that.
B
And
one
thing
to
note
is
that
this
is
very
similar
to
what
controllers
do
on
the
server
side,
like
they
read
resources
and
write
them,
but
it's
basically
about
changing
declarative
configuration
or
some
for
some
class
of
controllers,
and
so
this
may,
if
you're
noticing
neither
one
of
those
things.
This
is
a
good
place
to
look.
G
D
We
we've
moved,
as
we
mentioned
before,
we've
moved
quite
a
bit
of
code
to
staging
and
now
it's
actually
available
and
and
should
be
available
to
be
reused
as
libraries
and
so
we've
we've
already
refactored
made
pretty
significant
tree
factors
to
the
to
the
drain
code,
the
coop
cuddle
drain
code,
so
that
it
can
be
reused,
and
so
this
is
an
effort
to
to
start
the
ball
rolling
to
to
get
the
coop
cuddle
apply,
refactored,
so
that
it
can
be
reusable
so
initially.
In
fact,
you
know
before
this.
D
Just
just
please
bring
this
up
in
our
meetings
and
and
let's,
let's
actually
you
know,
we've
we've
got
the
opportunity
now
to
to
turn
this
code
into
reusable
libraries.
So,
let's,
let's
do
that
and
apply
especially,
is
one
of
the
biggest
things
that
we
have,
and
so
this
is
just
a
beginning.
In
order
to
make
this
code
more
reusable
thanks
and
if
you
get
any
questions
yeah
does
anybody
have
any
questions?
I.
H
D
A
H
H
A
With
or
with
regards
to
the
apply,
refactor
or,
most
importantly,
does
anyone
have
any
suggestions
which
other
parts
of
it
shuttle
should
be
refactored
I.
Remember
that
I
have
I,
have
a
big
fad
to
do
on
my
never-ending
to-do
list
to
refactor
overall
state
of
the
packages
inside
of
queue
cuddle
and
the
reason
for
that
is
over
time.
We
gain
some
unnecessarily
baggage
here
and
there
and
we
could
probably
try
to
straighten
a
few
things
along
the
way.
A
Some
of
the
changes
are
from
the
times
when
we
were
switching
from
internal
API
to
external
API,
so
you
might
encounter
packages.
They're
called
version
that
internal
and
similar,
if
I
get
the
time
and
hopefully
a
bit
closer
to
the
holidays
and
with
less
additional
work,
I
might
be
able
to
get
to
it
and
try
to
clean
a
few
a
few
bits
and
I
already
apologized
everyone
if
I
mess
your
PR
on
the
road
to
this,
because
on
my
crutch,
quite
a
bit
of
the
directories
on
the
way
to
to
clean
the
structure.
A
I
So
I
think
one
of
the
things
that
I
ran
into
is
I
just
I.
Don't
have
a
lot
of
domain
knowledge
yet
into
helping
some
of
these
things.
So
I
wanted
to
propose.
Maybe
getting
together,
like
you
know,
once
a
week
or
on
the
off
week,
where
we
don't
have
a
meeting
and
maybe
just
like
trying
to
go
through
issues
for
like
30
minutes
together,
I.
A
Mean
I'm
trying
to
cut
the
number
of
meetings
of
not
adding
additional
ones
and
from
the
past
I
know
that
there
are
some
meetings
where
we
don't
fill
in
the
entire
hour.
Can
we
start
by
devoting,
let's
say,
half
of
the
current
60
line
meetings
or
one
third,
let's
say
and
go
through
those
issues
to
start
with,
so
that
we
don't
multiply
the
meetings.
If
we
see
that
there
are
more
issues
or
something
like
that
or
we
prove
this
is
working
fine.
We
can
then
rethink
the
idea
of
having
a
separate
meeting
now.
B
A
B
D
C
D
B
A
We
can
go
and
discuss
through
several
of
them
and
then,
if
you
can
somehow
track
our
progress,
let's
call
it
that
way
so
that
we
know
which
word
triage
and
which
not.
That
will
definitely
help
us
untangle.
The
the
load
of
bug
because,
like
like
Sean,
said
and
I'm
as
well
as
guilty,
as
as
anyone
else
of
not
doing
this
promptly,.
I
A
A
E
B
That
was
that
better,
better
okay,
yeah
I,
just
like
I,
upgraded
Mac
and
now
it
wants
totally
different
permissions,
for
example,
okay.
So
what
I'm
going
to
demo
is
just
like
some
of
the
work
we've
been
doing
and
customized
so
specifically
a
set
command
for
like
the
equivalent
of
the
setters
and
control,
but
a
little
more
flexible.
B
So
there's
this
customized
config
command
with
the
number
of
things
in
here
and
I
have
I
just
cloned
the
examples
directory
from
jimeniz
examples,
specifically
the
khakhra,
CD,
subdirectory
and
so
I
can
run
customize
a
tree
on
this
thing
and
it
will
kind
of
print
out
the
contents
of
it.
So
for
reference.
This
is
like
the
configuration
in
here
and
so
I
can
ask
it
for
kind
of
a
view
that
has
specific
fields
print
it
out
under
what
and
there's
I
can
run
customize
on
this
directory
and.
B
It
will
show
me
this
one
setter
for
image
tag
which
you
can
see
is
listed
in
that
tree,
so
I
can
set
image
tag,
for
instance,
to
be
one
dot,
one
dot
three
and
then
as
well,
show
me
that
it's
been
changed
here
and
then
in
the
tree.
They'll
show
me
that
it's
been
changed
as
well,
and
the
way
this
has
been
implemented
is
by
a
comment
on
that
field
here,
and
so
it
basically
says:
here's
a
name
of
something:
that's
a
string.
B
The
current
value
is
this,
and
here's
like
a
possible
marker
to
used
for
doing
substitutions,
and
so
this
enables
that
field
to
be
substituted,
and
so,
if
I
do,
if
I
wanted
to
add
a
new
one,
for
instance,
I
wanted
to
have
the
port.
I
went
like
these
8080
ports
to
be
be
able.
Settable
I
could
do
something
like
Congress.
C
B
F
B
B
B
So
this
time
I'm
going
to
change
it
to
xgp
port
I'm,
going
to
change
by
288
one
and
then
maybe
I'll.
Add
a
I
can't
write
that
step
by
our
own
by.
B
You
can
support
both
like
partial
values
like
that
tag,
for
instance,
just
substituting
that
part
of
the
image
tag
ordering
the
whole
value
supports
the
person,
the
correct
type
and
it
supports
setting,
for
instance,
different
different
fields
and
different
resources.
So
you
can
have
a
setter,
for
you
know
your
food,
your
food
appointment
and
your
bar
deployment
separate
setters,
and
you
can
set
set
fields
together
right
in
this
case,
the
port's
field
container,
the
container
port
filled
in
the
staple
set
image
and
the
air
container
and
the
service
support
built
here.
B
So
that's
a
quick
demo
of
that
command.
Again,
these
are
explorations.
What
we're
trying
to
do
is
quickly
develop
prototypes
that
people
can
use
to
say
yes,
I,
like
this.
No
I,
don't
like
this
on
things
that
have
been
long-standing
problems
and
that
we
do
have
solutions
for
that
are
just
insufficient,
and
these
are.
These
are
actually
great
way
to
get
contributing.
If
you're
looking
for
areas,
you
can
jump
in
and
you
want
to
contribute
code
because,
because
their
alpha,
we
don't
have
to
spend
quite
as
much
time
debating
like
what
should
the
long-term
API.
B
H
H
So
I
guess
just
for
context:
hi
everyone,
my
name
is
Tyler
I've
been
using
customizing
production
on
a
vehicle,
telematics
platform
that
supports
roughly
a
million
vehicles
for
the
past
two
years
and
quickly
ran
into
lots
of
issues
around
variable
assignments,
and
things
like
that.
But
I'm
speaking
specifically
to
this
feature
edition,
like
looking
at
the
way
you've
modified.
H
H
B
Thanks
for
that
feedback,
this
is
not
to
solve
your
problem,
so
you
have
a
solution
that
works
for
you
right
and
so
there's
a
couple
things
I
think
to
key
on
one.
Is
some
people
want
a
CLI
to
see
what
options
they
can
set
and
then
set
them
from
some
field
right
or
someplace?
Right
so,
let's
say
I'm
getting
about
it
from
a
G
cloud:
mm-hmm!
That's
where
I
get
the
thing
I
would
have
to
send
that
into
the
declarative
file.
B
H
Letters
that
we
have
thus
far
are
based
on
the
structured
data
that
we
know
about
the
amyl
itself,
like
if
you're
saying
that
people
routinely
need
to
replace
container
ports.
I
agree
that
there
should
be
a
setter
method
for
that,
but
the
idea
that
it
is
implicitly
defined
in
the
amyl
itself
individually,
for
that
particular
manifest
and
could
potentially
be
different
in
every
single
manifest
throughout
the
Internet
to
me,
seems
like
missing
the
point
of
the
standardization
that's
available
through
structured
data.
H
So
if
it
turns
out
that,
like
container
ports
or
a
thing
that
need
to
be
modified
routinely,
wouldn't
it
make
more
sense
to
make
that
a
standard
that
applies
to
every
single
manifest
and
every
single
possible
object.
That
itself
has
a
container
port.
So
there's
a
standard
way
to
do
that,
no
matter
where
it
came
from.
Maybe.
B
H
Agree
I'm
only
speaking
to
the
basic
I'm
speaking
to
the
basic
tenant,
that's
broadcast
on
the
main
page
of
customers,
I
customized
and
has
been
a
party
line,
that's
been
drawn
for
the
last
two
years,
which
is
that
inline
templating
is
bad
and
wrong
and
we're
trying
to
solve
that
without
doing
that,
it's
not
the
case
anymore.
Well,.
B
It's
important
to
understand
a
couple
things:
okay,
this
is
not
so
building.
We
have
to
look
at
the
CLI
surface
for
Stickley
mm-hmm
right,
so
in
customize
you
are
in
customized,
build
in
the
customization
UML.
That
is
still
the
party
line
yeah
right.
Okay,
now
generally,
why
do
we
say?
Templating
is
bad
in
substitution?
There's
a
reason.
It's
not
just
like
Dogma
right,
I,
agree
and
the
reason
is
that
tools
cannot.
If
you
template
something
tools
cannot
analyze
it.
They
cannot
look
and
say.
Is
this
value
good
right?
They
cannot.
B
Can
a
tool
look
at
this
file
and
do
static
analysis
on
it
and,
yes,
it
can
write
the
tool,
can
look
this.
It
will
parse.
You
can
run
to
bail
against
this.
You
know
later
validate
or
it
won't
write
and
if
they
don't
modify
it
and,
let's
say
I,
run
the
substitution
command
and
I'm
like
I'm,
pulling
a
value
from
some
other
source
right.
That
I
would
otherwise
have
to
set
into
the
file
right
in
this
case,
like
I,
just
create
I,
I'm
part
of
the
legacy
system.
B
It
spits
out
some
value,
which
is
the
new
release
thing
that
gets
applied
as
a
label
and
I
want
to
change
that
particular
label.
That
has
my
release
that
I
get
from
someplace
else.
This
actually
provides
a
structured
way
that
adheres
more
closely
to
those
principles,
because
this
is
actually
attaching
metadata
to
that
field.
This
is
something
that
would
be
an
open
API
first
and
it's
like
open
API
could
have
this
same
data.
B
The
problem
is
it's
hard
to
package
that
schema
data,
open
API
with
the
file
with
the
configuration
itself
right,
so
I
just
think.
Instead
of
this
less
like
a
templating,
arbitrary
sub
solution
perspective
and
more
like
what?
How
can
we
get
merge,
the
sort
of
data
that
we
would
want
to
get
from
the
schema
and
open
a
guy
into
the
configuration
right
and
there's
some
so
that
open,
API,
wouldn't
necessarily
even
there's
some
stuff?
We
need
to
figure
out
like
how
does
open
API
allow
you
like.
B
So
you
can,
for
instance,
want
to
be
very
possible
to
say,
I
want
to
substitute
this
value
into
this
particular
resource,
but
not
that
other
resource,
but
when
I
substituted
it
into
this
one
resource
it
has
to
like
there's
a
reference
to
this
other
field,
and
that
one
has
to
be
updated
great.
So
you
can
have
these
complex
relationships
that
needs
to
be
embedded
in
the
schema
as
well.
Yeah.
H
H
I
agree
that
there
I
totally
agree
with
the
idea
of
implicit
modification
versus
declarative
modification
and
supporting
both
in
a
holistic
way,
I
think
just
speaking
specifically
to
this
implementation,
maybe
because
it's
alpha
now
isn't
the
time
to
talk
about
it,
but
like
you're,
calling
this
thing
HTTP
port,
which
is
duplicating
the
fact
that
it's
already
named
HTTP
and
it's
already
under
a
field
called
HTTP
port.
Like
this.
F
H
Me
just
looks
like
personally
from
my
experience,
like
a
maintenance
nightmare,
that's
going
to
sprawl
out
into
everything
to
solve
a
problem,
that's
like
on
the
underlying
replacement
that
you're
trying
to
make
there's
already
a
functioning
solution
for
that
doesn't
require
this,
that
you
could
expose
in
the
CLI
and
I
hear
you
saying
like
go
implement
that
and
that
makes
perfect
sense
to
me
you're
the
one
working
on
it.
You
do
what
you're
going
to
do.
H
I
will
say
that
for
me
personally,
I'm
not
highly
incentivized
to
contribute
to
customize,
because
I
spent
about
three
months
with
another
main
another
contributor
trying
to
get
a
feature
in
that
was
ultimately
just
closed
and
ignored,
and
not
reviewed
so
well.
I
do
have
the
capability
to
go
and
provide
an
implementation.
I
don't
have
the
belief
that
this
project
is
going
to
support
it.
You've
historically
held
an
extremely
hard
line
on
not
doing
exactly
this
kind
of
thing.
H
So
it's
a
bit
like
head
Spinney
for
me
that
suddenly
there's
ten
people
working
on
this
project,
adding
all
kinds
of
functionality
to
it
that
frankly
it's
just
it
seems
like
it's
incomplete
odds
with
what
the
project
has
been
holding
a
hard
line
on
for
the
last
two
years.
So
for
me,
it's
new
I'm
glad
that
there
are
additional
people
working
on
it.
I
think
this
implementation
is
bad.
I
use.
It
I
understand
that,
like
I
understand
what
you're
trying
to
do,
I
think
this
there's
a
solution
that
doesn't
necessitate
these
comments.
But
okay.
B
B
B
H
B
Ii
or
the
CRT
version
right
for
series
that
are
installed
right
and
so,
which
version
of
the
open
API
do
you
use
where
do
I
get
it?
Does
this
mean
I
have
to
have
a
cluster
with
that
CRD
installed
in
order
for
me
to
work
with
this
resource,
or
is
it
downloaded
as
a
side
channel,
in
which
case
my
tooling
has
to
read
some
open
API
file,
that's
packaged
with
this
resource.
That
is,
you
know,
so,
like
there's
a
lot
of
challenges
with
having
a
separate
schema
right
that
there's
no
clear
way
of
getting
to.
B
So
one
like
thing
to
look
at
that's
related
to
this
is
the
queue
language
I
know
if
you
see
that
it's
around
unification
and
it
unifies
schema
and
data,
so
the
data
itself
has
the
schema
defined
through
essentially
constraints
and
these
constraints
to
unified,
and
then
so
you
go
from
schema
to
data,
as
you
add,
more
restrictive
constraints
in
it,
and
so
this
is
along
that
same
vein
of
thinking
of
this
is
effectively.
It
comes
like
similar
in
that
notion,
like
there's
constraint,
ish
stuff
here,
like
the
type
for
instance,
is
on
there.
H
The
total
sense
to
me
and
as
structured
data
as
like,
an
annotation
as
opposed
to
a
comment
that
the
set
command
could
add
to
the
resource
but
again
I
feel
like
potentially
we're
getting
into
the
weeds
here,
but
I
see
where
you're
going
with
it
I
understand,
there's
value
and
who,
knowing
who
controls?
What
and
who's
made
changes.
I
just
think
that
comments
are
not
the
solution.
Cool.
A
Other
topics,
questions
I-
would
want
to
cut
this
one
at
this
point
in
time
since
we're
almost
at
the
top
of
the
hour.
I
want
to
leave
the
minute
for
the
stand
up
and
I
see
that
Eddie
put
together
already
four
issues
and
the
Bucks
Krupp
section,
so
I'll
probably
want
to
get
to
that
one
as
well.
For
setting
up
I,
see
customized
Kay
young
I'm,
not
sure
if
that's
fill
you
or
someone
else.
What
I
quickly
cover
that
one?
Oh
I.
B
A
A
Of
it,
I'll
just
go
with
the
ones
that
you
put
together
so
that
you,
first
of
all,
get
the
idea
of
what
I'm
looking
most
frequently,
how
I'm
debugging
stuff
and
that
that
should
help
you
get
the
ball
rolling
and
we
can
continue
on
the
next
meetings.
Does
that
make
sense,
yeah
works
for
me,
okay
cool,
so
the
first
one
you
put
together
is
about
cube
Caudill
label,
which
was
not
removing
a
label
from
a
node.
A
So
in
this
particular
case,
is
if
a
command
is
not
working
as
it
as
this,
as
it
is
expected.
The
first
thing
that
I'm
personally
asking
the
the
reporter
to
do.
We
can
ask
them
for
a
cute
cuddly
comment
with
an
increased
verbosity
level.
If
you
ask
for
a
positive
levels,
7/8
or
up
to
9
you're
gonna,
get
I,
can't
remember
the
exact
number,
but
it's
written
somewhere
down.
A
9
gives
you
everything
up
to
the
point
where
you're
getting
curl
commands
and
the
request
and
the
response.
The
lower
levels
are
cutting
some
data,
so
I'm
usually
asking
for
the
third
for
the
V
9,
and
then
you
can
see
an
actual
request.
That's
being
sent
over
to
the
server
which,
in
the
case
of
a
label
command,
it
will
be
a
a
patch
command,
which
is
what
basically,
you
did
and
ask
the
person
to
to
respond.
A
I
A
A
I
do
remember
that
at
some
point
in
time,
in
the
past
we
had
some
issues
with
nulls
and
the
api's
and
that
they
were
not
handled
properly.
It
might
be
one
of
those
cases
so
either
he
can
try
a
newer
version
and
see
if
the
problem
persists
or
or
eventually
we
can
go
back,
although
usually
we
try
to
fix
stuff
in
the
newer
version
so.
I
I
A
A
A
A
He
read
the
labels
once
again,
it
looks
like
there's.
There
might
be
some
additional
process
that
might
be
modifying
the
labels
afterwards,
because
the
the
response
from
the
server,
if
you
look
carefully,
does
not
have
the
label
set,
which
means
the
patch
was
put
together
correctly,
and
the
response
from
the
server
clearly
indicates
that
it
was
a
flight
on
the
server.
A
A
A
A
I
A
Yeah,
if,
if
it's
just
a
misunderstanding,
how
he
should
inject
the
lavash,
then
I'm
perfectly
fine,
with
with
closing
it.
If
you
see
that
the
actual
output
of
this
is
coming
from,
you've
got
old
completion
command,
then
we
might
want
to
fix
the
the
actual
template
so
that
it
does
produce
a
valid
output.
A
A
A
A
It
should
get
removed
only
if
it
fails,
but
not
by
default.
So
something
has
happened
in
between
here.
It
would
be
definitely
of
warranty
to
try
ask
for
that.
I
should
be
output
and
gets
what
happens
in
between,
although,
if
I
understand
correctly,
this
is
happening
infrequently
and
you
know
in
the
command
so
that
that
will
be
the
the
approach
that
will
take.
A
The
role
check
that
we
have
like
I
remember
the
exact
resource
that
is
being
created
when
you
pass.
The
name
is
not
constructed
correctly
or
it
is
constructed
so
that
the
server
gives
us
a
wrong
answer
will
be
valuable
to
get
the
output
and
if
it's
something
that
is
failing
in
the
current
is
happening
in
the
current
master,
it
should
be
a
pretty
hopefully
easy
fix
for
someone
interested
in
in
helping
with
with
the
cube
Caudill.